What Happens the Game After a Marathon Extra-Inning Game?
Last Thursday, baseball got weird and the Mets and Marlins played past midnight. After Travis d’Arnaud hit the go-ahead homer in the 16th, the catcher slowly trotted around the bases, admitting afterwards that he needed the invigorating effects of that moment just to complete the task. “The emotions of the home run helped lift my legs a little bit,” he said to James Wagner after the game regarding his tired knees. After the dust had settled and all the exhausted quotes were collected, though, the teams had to play another game later that day. What sort of effect would the marathon game have on that game?
Intuitively, you might expect the teams to have trouble scoring runs the next day. Tired legs, tired minds, tired bats, you’d think. Turns out that instinct is accurate… sort of.
With the help of my colleague Jeff Zimmerman, we took a look at all 4,117 games that have gone longer than 12 innings since 1974. As a barometer for offense, we used runs scored per nine innings played for each team. Here’s how the teams scored generally, both in the marathon game and in the game after the marathon game:
| Situation | Seasonal | Long Game | Game After |
|---|---|---|---|
| Runs Per Nine IP | 4.46 | 3.18 | 4.24 |
n=4117 games that went longer than 12 innings.
The most pronounced difference in run-scoring actually occurs in the long game itself. That makes a ton of sense to anyone who has watched batters swing away in extra innings, desperate to finish off the contest, and yet playing right into the pitchers’ hands by being so aggressive. It seems impossible to score that final run once you get to the 13th inning, and it looks like it’s possible the fatigue creates an in-game depression of run-scoring.
But the game after? It’s a little down from season averages. It might be surprising that it’s not down even further, though. In the case of Thursday’s game, for example, Mets sat Yoenis Cespedes and the Marlins sat Giancarlo Stanton, and everyone else was super tired. You’d think that would lead to fewer runs scored.
However, many of these games are played against the same team that took just part in the long game. Though it didn’t play out this way in the Mets and Marlins game on Friday — those two teams combined for seven innings of one-run relief the day after they pitched 24.1 innings combined — post-marathon-game bullpens are usually taxed, and they may help the other tired offense score more runs.
You’d think that the most pronounced effect might be a lack of extra inning games the day after a marathon game, then. Either the offense can’t score enough to tie the other team or the bullpen just can’t keep the lid closed, right? Teams have combined to go to extra innings 8.9% of the time after a 12-plus inning game. They went to extra innings 9.0% of the time, period, over the same time frame.
Back to Travis d’Arnaud’s exhausted trot for a second. It’s interesting to note that neither catcher who caught Thursday’s 16-inning epic played again the next day. The teams tried not to use relievers in both games — the highest number of batters any Thursday reliever saw on Friday was five. David Phelps faced five batters in both games, which doesn’t seem all that taxing in the larger scheme of things. A few stars got a day off to keep them healthy, maybe.
Long games make baseball players tired — more tired than a regular nine-inning affair. There can’t really be a doubt about that. But baseball itself is a grind, a day-in, day-out marathon that wears everyone down. So the players have learned to cope — by drinking lots of water and sleeping as well as they can, getting massages, using ice baths, downing Advil, whatever it is.
In a similar way, teams themselves have added a few coping mechanisms when it comes to long games — sit the catcher, avoid the relievers that pitched the most, sit a star — that have kept them from suffering terribly in the game after a marathon.
With a phone full of pictures of pitchers' fingers, strange beers, and his two toddler sons, Eno Sarris can be found at the ballpark or a brewery most days. Read him here, writing about the A's or Giants at The Athletic, or about beer at October. Follow him on Twitter @enosarris if you can handle the sandwiches and inanity.
Wouldn’t the reduced run-scoring in a marathon game just be a result of tie games occurring more often at lower scores? I think that would have a bigger correlation than player fatigue. You rarely get 5-5 extra inning games whereas 1-1 or 2-2 games heading into extras is comparatively common.
This, plus it’s comparatively rare for an away team to score (x>0) runs in the top of an extra inning followed by the home team scoring exactly that many runs in the bottom of the inning, meaning that, if you’re selecting for games that last at least 12, the proportion of 0-0 tenth, eleventh, and twelfth innings will be greater than average (because the tenth, eleventh, and twelfth innings that feature scoring usually end the game).
That’s not to say there’s not an effect, but I think a large portion of the difference is in the selection and not the gameplay itself.
I doubt there is any reduced run scoring in marathon games. The total amount of runs scored is probably higher on average for the entire game. Eno calculated Runs/9IP which would probably be lower though just because in order to even get to a marathon game there is most likely no scoring from innings 10 through the 1 inning before the game ends
Excellent point that I didn’t anticipate! Thanks for pointing this out! The distribution of scores suggests that yes, they will be lower-scoring games on average.
Interesting article. But being from Boston I assumed this was a post about the games following the annual 11 o’clock marathon Monday games. Which are probably no different from other games.
Thought this same exact thing. Mostly due to the timing of this article relative to the Marathon, and the subsequent, “game after a marathon,” taking place today.
Is there a difference in run scoring when the two tired teams face off against different (and theoretically fresher) opponents the day after their marathon game?
“The most pronounced difference in run-scoring actually occurs in the long game itself. That makes a ton of sense to anyone who has watched batters swing away in extra innings, desperate to finish off the contest, and yet playing right into the pitchers’ hands by being so aggressive. It seems impossible to score that final run once you get to the 13th inning, and it looks like the fatigue creates an in-game depression of run-scoring.”
This isn’t the most important piece of the article, since you’re writing about what happens the next day after the marathon game, but I have an issue with the interpretation of the data regarding what happens in the marathon game itself. Correlation does not imply causation.
Fewer runs are scored in long games. This makes a ton of sense… but for very different reasons:
1) Extra inning games should be lower scoring on average, because any game that goes to extra innings must have been tied after nine innings, and lower scoring games are much more likely to be tied after nine innings. This is for the same reason that you’re much more likely to be able to flip two coins and have them come up the same way than you are to get two dice to come up the same way (HH, TT, TH, HT – there’s a 50% chance of a “tie,” vs dice in which there’s a 1 in 6 chance).
Let’s say I told you that one of the following two games went to extra innings and told you to guess which one, but only gave you one bit of information: one game had 4 runs scored in nine innings, the other game had 12 runs scored in nine innings. The smart choice is the 4 run game, as the only possible outcomes are 4-0, 3-1, 2-2, 1-3, 0-4, meaning that one of five possible outcomes results in a tie. For the 12 run game there are thirteen possible outcomes, and only one, 6-6, leads to extra innings. The fewer runs scored in the game, the more likely the game is to go to extra innings. So fewer runs is a cause, not a consequence, of the game going to extra innings.
2) This effect continues into extra innings as well. The most likely outcome of any frame in baseball is 0 runs (plurality, if not majority). Since any extra inning that leads to the continuation of the game must have involved equal frames from both teams, the best chance of this occurring is for both teams to score 0 runs. The more runs Team A scores in the top of the inning, the less likely it becomes that Team B matches them exactly. So by selecting only games that went >12 innings, you’re already selecting a huge number of games in which innings 10, 11, and 12 involved 0 runs scored (not all, of course, but I’d wager that the majority of that 4117 game sample involved no runs in innings 10-12). If you take a sample of 9 inning games that had an average of 4.46 runs scored per 9 innings, then you tack on 3 scoreless innings to each game, the runs per 9 drops to 3.35. Combined with my first point, I’m guessing this just about explains the 3.18 runs per nine that you found.
So a “long game” is more likely to have fewer runs scored per nine innings, yes. But it’s not that the long game is causing fewer runs to score, but rather that fewer runs scored makes the game more likely to go long. By selecting games that went >12 innings you’re selecting games in which fewer runs were scored in innings 1-12.
It’s very possible that the perceived effect of it seeming “impossible to score that final run once you get to the 13th inning” does exist. But to support that claim you’d need to show different data. Analyze how many runs are scored per 9 innings *from inning 13 onward* and if the finding stands, then you can make that claim.
Speaking of “marathons”… there were no comments when I started typing up this colossus, but of course after posting it it seems DayNife and jruby have already said what I was trying to say far more succinctly.
Anyway, I want to clarify that my post is not meant to take away from the main finding of the post, however, which was regarding what happens the day *after* a long game. The fact that teams scored nearly a quarter of a run fewer per nine innings after a long game seems very significant and interesting, and I fully agree with both of your explanations for why that might be: that hitters are tired (or being rested) so offenses are less likely to score, but that tired bullpens might be mitigating that outcome somewhat.
In fact, my first point above might actually help explain why the data didn’t support your hypothesis that day-after-long-game games weren’t significantly less likely to go extra innings. In fact, they probably *are* a little less likely to go extra innings when you control for the fact that these are lower scoring games to begin with, since lower scoring games are more likely to go to extra innings.
If the game is going long, would a team send the next day’s starter home (or to the hotel) to sleep on a normal schedule? I personally would tell the starter and any other players that were taken out of the game to get normal sleep.
But then the longer into extra innings you go, the more likely you’re about to need the next day’s starter right then.